Maarten Sap

Maarten Sap is an Assistant Professor at Carnegie Mellon University's Language Technologies Institute, with a courtesy appointment in the Human-Computer Interaction Institute. He also serves as a part-time Visiting Research Scientist & AI Safety Lead at the Allen Institute for AI (2024–present).

• PhD in Computer Science & Engineering (2015–2022) from University of Washington, advised by Noah Smith and Yejin Choi
• BSc in Communications and Information Systems (2010–2014) from École Polytechnique Fédérale de Lausanne

His research focuses on socially intelligent AI systems through three core areas: (1) measuring and improving AI's social/interactional intelligence, (2) assessing and mitigating socio-cultural biases in language models, and (3) building narrative technologies for prosocial outcomes. Key themes include AI ethics, toxicity detection, cultural adaptability, and human-AI collaboration.

Recent publications highlight his work on safety frameworks like HAICOSYSTEM, clinical reasoning alignment (ALFA), multilingual moderation (PolyGuard), and social intelligence evaluation (SOTOPIA). He received the 2025 Okawa Research Grant and NAACL 2025 Best Paper Runner-Up award. His lab advises 10 PhD/MLT students across CMU and MIT, with co-advising relationships at UW and EPFL.

• Scientific Awards:
  • 2025 Okawa Research Grant (7 US recipients)
  • NAACL 2025 Best Paper Runner-Up
  • EMNLP 2023 Outstanding Paper
  • FAccT 2023 Best Paper
• Advising:
  • Primary advisor to Joel Mire, Mingqian Zheng, Jimin Mun, and Akhila Yerukola
  • Co-advisor to Dan Chechelnitsky, Karina Halevy, Jocelyn Shen, and Xuhui Zhou
• Leadership:
  • Co-organizer of Agent Workshop @ CMU (2025)
  • Advisory board member for Socially Responsible Language Modeling (NeurIPS 2024)
  • Senior Program Committee - ACL (2020–present)